Local AI Image Classifier

Classify images by description — no cloud API needed, everything runs locally.

How it works

Uses OpenAI CLIP to compute similarity between an image and a list of text labels. The highest-scoring label wins.

from PIL import Image
import clip, torch

model, preprocess = clip.load("ViT-B/32")
image = preprocess(Image.open("photo.jpg")).unsqueeze(0)
labels = ["a cat", "a dog", "a car", "a tree"]
text = clip.tokenize(labels)

with torch.no_grad():
    logits, _ = model(image, text)
    probs = logits.softmax(dim=-1)

best = labels[probs.argmax()]
print(f"Predicted: {best}")

Interface

Flask-based web UI. Drop an image, enter labels, get a classification.

Status

v0.1.0 — proof of concept. Works on CPU, slow on large batches.